We previously developed a vaccine based on coimmunization of OVA peptide (aa323–339 restricted by I-A^d, recognized by DO11.10 TCR) with a DNA construct carrying the same epitope coding sequence. The immunization induced antigen-specific CD4⁺CD25⁻ asTregs and prevented the host from OVA sensitization-induced asthma (33). To evaluate the therapeutic value of this protocol in established asthma, we induced airway inflammation in BALB/c mice by multi-step sensitization/challenging of intranasal injection of OVA (Figure 1A). The symptomatic mice were then treated with coimmunization of OVA + pVAX1-OVA (Co-OVA) per schedule described in Figure 1A, with antigen-mismatched control treatment of Der p 1 + pVAX1-Der-p1 (Dermatophagoides pteronyssinus peptidase 1, Co-DERP1). In the asthmatic mice, respiratory resistance and T cell infiltration in the lung were reduced following the Co-OVA treatment (Figures 1B–D), in comparison with the limited effect of PBS sham or Co-DERP1 control treatment. Histology study showed that inflammatory reactions in the lung were evident in the sham and control treatments, while the Co-OVA ameliorated the inflammatory infiltration (Figures 1E,F).

In Co-OVA-immunized Foxp3-eGFP transgenic mice without asthma induction (scheme in Figure S1A in Supplementary Material), a subset of CD25⁻Foxp3⁺ T cells appeared in CD4⁺CD25⁻ population (9% versus 0.9%) while the size of CD25⁺FoxP3⁺ tTreg population barely changed, suggesting that the Co-OVA protocol preferentially led to CD4⁺ T cell conversion to asTregs, without recruiting or expanding tTreg pool on site (Figures S1B,C in Supplementary Material). As a systematic control for the Co-OVA, Co-DERP1 immunization showed similar induction in the CD4⁺CD25⁻ population rather than the CD25⁺FoxP3⁺ tTreg population (Figure S1C in Supplementary Material). Functionally, the Co-OVA asTregs sorted to homogeneity were able to suppress DO11.10 T cell proliferation in response to OVA stimulation in vitro (Figures S1D–F in Supplementary Material). Although any activated T cells or Tregs would compete with the naïve T cells for antigen and available IL-2, as both of them express CD25 and might be antigen-specific, the near absence of any DO11.10 T cell proliferation still suggested an active suppression event.

To study whether those asTregs responded to the antigen specifically, they were stimulated with OVA or Der p1-loaded APCs in vitro (Figure 2A). Co-OVA asTregs upon OVA but not control Der p1 stimulation secreted IL-10, TGF-β, and IL-35 (Figure 2B; Figure S1G in Supplementary Material) and showed robust proliferation (Figure 2C; Figure S1H in Supplementary Material). Those functions were entirely dependent on MHC class II (Figures 2B,C). In a mirror image, Co-DERP1 asTregs responded to Der p1 rather than OVA with a similar spectrum of activations (Figures 2D,E). This proliferation upon cognate antigen stimulation appeared to be substantial in comparison with a more strained antigen-driven division of tTregs reported in some articles (34–37). Our results therefore suggest that asTregs may rise in the presence of overt, specific antigens.

To rule out additional in-host changes that may act as confounding factors in parallel to asTreg induction, we produced Co-OVA or Co-DERP1 asTregs via the coimmunization protocol, isolated and transferred them into OVA or Der p 1-sensitized asthmatic mice, thus leaving antigen specificity matching/mismatching between effector T cells and asTregs the only variable in the system (two-step schemes are in Figures 2F,G). OVA asthmatic mice receiving Co-OVA asTreg transfer rather than Co-DERP1 asTreg showed amelioration both in plethysmography readouts and in T cell infiltration. Similar reductions in Der p1 asthmatic mice were only evident with Co-DERP1 asTregs (Figures 2H,I). These data collectively demonstrate that asTregs induced by coimmunization inhibit T cell responses and inflammation in asthma in an antigen-specific manner.

Although there is agreement that Treg-mediated suppression requires close proximity, either to Teff (38–42) or DCs (43, 44), it is less clear where the primary location of such a suppressive function is carried out. We transferred GFP⁺Co-OVA asTregs into recipient OVA-sensitized asthmatic mice and examined the lung and its associated lymphoid organs for GFP⁺ cells 2 days after their infusion. Unexpectedly, measured by total cell numbers recoverable, most of the asTregs were located in the lung-associated hLN, limited numbers of asTregs were found in the auxiliary LN and they were essentially undetectable in the lung (Figure 3A). Visual inspection also revealed an expanded size of the hLN (Figure S2A in Supplementary Material). The exceptionally high recovery of these cells in hLN might be influenced by the prior imprint priming in the donor and their expansion upon antigen exposure in the recipients. We did not pursue these possibilities further. Histology of hLN sections showed that the asTregs distributed in both B cell and non-B cell zones (Figure S2B in Supplementary Material). To follow their dynamic changes in hLN, numbers of GFP⁺ asTregs were continuously monitored over a period of 2 weeks. In our system, both specific and non-specific asTregs trafficked into hLN on day 2. However in hLN, Co-OVA asTregs increased by days 7 and 14 (Figure 3B), whereas Co-DERP1 asTregs faded gradually, suggesting an antigen-driven retention of OVA-induced asTregs.

Draining LNs are the primary location of antigen-specific T cell priming (45, 46). Maternal blocking of lymphotoxin ablates secondary lymphoid organs in the offspring (47). To demonstrate the role of hLN for asTregs to exert their suppression, LN-null mice were generated by maternal injection of LTβR-Ig during gestation. After asthma induction in these LN-null mice, asTregs were adoptively transferred. hLN in mice born following the maternal injection of LTβR-Ig was not visible (Figure S2C in Supplementary Material). Yet, the absence of hLN did not affect the asthma establishment (Figures 3C,D). However, in these mice, infusion of Co-OVA asTregs was no longer able to suppress asthma, measured by both airway restriction and CD4⁺ T cell infiltration into the lung (Figures 3C,D). These results suggest that in our model, the anatomic structure of LNs in general, particularly hLN, is essential for the suppression of T cells; the mere provision of the full complement of asTregs is not sufficient.

Lymphocyte ingress and egress with respect to LNs are regulated by chemotactic factors (48–50). The same regulation is also exerted on Tregs (51). To identify any surface molecules that might be responsible for the retention of asTreg, we compared the expression profile of chemokine receptors and adhesion molecules on the antigen-matched asTregs recovered from hLN 2 days after their transfer with those of the mismatched control (Figure 4A). Ingress receptors on Co-OVA asTregs, including CCR4, CCR6, CCR7, CCR8, CCR9, and CXCR3 were found to be expressed at levels similar to those on Co-DERP1 asTregs; nor was any difference detected for adhesion molecules VLA-4, CD44, CD69, and CD62L (Figure 4B). Interestingly, both populations showed no detectable CCR3 and CCR5. However, the expression dynamics of egress chemokine receptor S1p1 on the Co-OVA asTregs showed a precipitous drop after the transfer, in a sharp comparison with the steady expression on Co-DERP1 asTregs (Figure 4C). Antigen-triggered S1p1 regulation helps T cells to localize in the peripheral lymphoid organs (52), however whether it holds true for asTregs in vivo is not known. To confirm this central hypothesis, donor Co-OVA asTregs were cocultured with antigen-loaded APCs in vitro. As depicted in Figure 4D, expression of S1p1 on the asTregs was significantly reduced only in the culture with OVA-loaded APCs. This reduction was blocked by the addition of anti-MHC-II mAb, suggesting the class II antigen presentation was essential to induce the S1p1-dependent retention of those asTregs.

Transwell experiments in vitro with donor Co-OVA asTregs before transfer, and those hLN Co-OVA and Co-DERP1 asTregs recovered day 7 after their transfer into OVA-sensitized mice (without endogenous induction of asTregs by DNA/antigen coimmunization) (Figure 4E), also confirmed that antigen-driven S1p1 downregulation on specific Co-OVA asTregs impaired their chemotactic capability under a S1p gradient (Figure 4F). These results suggest that asTregs, upon second stimulation by the same antigen in hLN, downregulated their surface S1p1 and no longer heeded the call of S1P in the periphery; and asTregs activated otherwise without this second antigenic hit were not restrained in hLN. To verify that S1p1 reduction on the asTregs was required for the asthma amelioration, we overexpressed S1p1 in asTregs (designated as S1p1^ov asTreg) (Figure 4G; Figures S3A,B in Supplementary Material: QC of S1p1 overexpression). After adoptive transfer, the S1p1^ov asTregs were no longer retained in hLN (Figure 4H) and failed to suppress asthma (Figure 4I). Therefore, the antigen-triggered S1p1 reduction on the asTregs is crucial for its accumulations in hLN and asthma treatment.

Asthmatic remodeling of the airways is a chronic process: persistent priming of T cells is a crucial underlying cause (53, 54). We analyzed whether Co-OVA asTregs limited airway inflammation by reducing T cell priming in hLN. OVA-sensitized, asTreg-recipient asthmatic mice were infused with cognate DO11.10 CD4⁺ T cells on day 7 after the asTreg transfer, and the maturation and polarization of these T cells in hLN were analyzed 3 days later (Figure 5A). Most KJ1-26 (DO11.10 TCR-clonal) donor cells retained their CD62L⁺CD44⁻ naïve state in Co-OVA asTreg-treated recipients (Figure 5B) and were CD25 and CD69 negative (Figures 5C,D), confirming their resting state. It should be noted that any activated T cells or Tregs might contribute to the suppression by competition for antigen and IL-2 in vivo, as was seen in the in vitro assay, however the high percentage of resting T cells in the Co-OVA group was indicative of an active suppression. In addition, Th2 polarization of donor cells in hLN was also controlled by specific asTreg treatment as a two- to threefold reduction in the number of IL-4⁺ KJ1-26 cells was seen (Figure 5E). A small (~1.5%) yet statistically significant increase of IL-10⁺ donor T cells was evident in the same treatment group (Figure 5F), collectively indicating a switch to a less pro-inflammatory state in these T cells. Both PBS and non-specific Co-DERP1 asTreg recipients showed activation and polarization of donor DO11.10 T cells (Figures 5B–F), suggesting the priming controlled by Co-OVA asTregs was antigen dependent. These results support the notion that antigen-specific naïve T cell priming is controlled in hLN in the presence of antigen-specific asTregs.

As hLN is the main location of asTreg-mediated suppression, we were curious if effector T cells were also attracted to this location for optimal inhibition, which would also limit the number of those T cells in the lung. DO11.10 CD4⁺ T cells from sensitized mice as Teff were i.t. (intratracheal) transferred into the asTreg-treated OVA asthma model 7 days after the asTreg transfer (Figure 6A; Figure S4A in Supplementary Material). As the baseline without asTreg treatment, within 24 h, half of donor inflammatory T cells infiltrated lung tissues which cannot be washed out as bronchoalveolar lavage (Figure S4B in Supplementary Material). Recruitment of lung-infiltrating DO11.10 Teffs back to recipients’ hLN was detected at 36 h. The number and percentage of DO11.10 Teffs in hLN were increased in allergen-specific Co-OVA asTreg-transferred group (Figures 6B,C). Non-specific asTreg-treated recipient mice showed a lower T cell recruitment to hLN similar to the sham-treated. Importantly this accumulation of Teff also relied on the antigenic match between the suppressor asTregs and the suppressed Teffs.

We analyzed the donor DO11.10 cells’ apoptosis and proliferation, as these factors affected the output of these cells from hLN. As shown in Figure 6, Teffs were less proliferative as measured by Ki67 staining (Figure 6D) and had higher levels of apoptosis (Figure 6E) in the presence of Co-OVA asTregs.

Mirroring the asTregs, expression profile of ingress chemokine receptors, adhesion molecules and egress chemokine receptor S1p1 on DO11.10 Teffs recovered from the hLN on day 10 was analyzed (Figure 7A). No statistical difference was observed except for significantly increased CXCR3 and VLA-4, as well as a decreased S1p1 (Figure S5 in Supplementary Material and Figure 7B). To verify the role of these molecules in Teff trafficking through hLN in the presence of asTregs, donor DO11.10 Teffs were premodified with CXCR3 and VLA-4-specific short harprin RNA (shRNA) or the vector with S1p1-overexpressing sequence (Figures S3C–E in Supplementary Material). Modified donor cells were transferred and analyzed in parallel experiments (Figures 7C,D). While neither sham nor Co-DERP1 immunized retained Teffs, forced expression of S1p1 enabled the egress of Teffs from hLN (Figure 7D). CXCR3 and VLA-4 knockdowns showed a marginal decrease of Teff retention (Figure 7D). Therefore, while Teff egress, similar to Treg, is mostly regulated by S1p1, they may be additionally controlled by enhanced retention via CXCR3 and VLA-4. These observations suggest that these receptors are coordinated for the eventual entrapment of Teffs in hLN.

Together, we identified an antigen-specific S1p1 downregulation on Tregs and effector T cells, preventing their egress from hilar LNs in an allergen-induced asthma model, thereby enabling long-lasting and efficient suppression of effector T cells at this site. This regulatory event in the LN is critical to the amelioration of airway inflammation in the lung.

Foxp3-eGFP/C mice were purchased from JAX. DO11.10 mice were kindly provided by Dr. Minghui Zhang (Tsinghua University, Beijing, China). BALB/c mice were purchased from Vitalriver (Beijing, China). All mice were bred and maintained under SPF conditions at the central animal facility of Shanghai Medical College (SHMC). All protocols of animal experiments are approved by the Committee of Experiment Animals of SHMC.

DNA vaccines (pVAX1-OVA and pVAX1-Der-p1) were produced previously (82) and prepared by EndoFree Plasmid Maxi Kit (Qiagen). His-tagged recombinant Der p 1 protein was prepared by Ni-NTA agarose (Qiagen) and Detoxi-Gel Endotoxin Removing Gel (Thermo). Endotoxin of each vaccine was below 10 EU/mg. All reagents were from Sigma-Aldrich unless specified otherwise. Magnetic CD4⁺ T cell purification kit was purchased from R&D System. LTβR-Ig protein was a kind gift from Dr. Yangxin Fu (University of Chicago) and Dr. Mingzhao Zhu (Institute of Biophysics, Chinese Academy of Science). Lentiviral systems of S1p1 overexpression, CXCR3, and VLA-4 knockdown were gifts from Dr. Ying Wan (Third Military Medical University, Chongqing, China). Antibodies for CCR5 (7A4), CCR7 (4B12), CD44 (IM7), CD62L (MEL-14), CD4 (GK1.5), CD25 (PC61.5), CD69 (H1.IF3), DO11.10 TCR (KJ1-26), IL-35/IL-12p35 (4D10p35), B220 (RA3-6B2), and MHC-II (M5/114.15.2), as well as all ELISA kits, were purchased from eBioscience. Antibodies for CCR3 (TG14), CXCR3 (CXCR3-173), CCR6 (29-2L17), CCR9 (9B6), CD11a (M17/4), and CD49d (R1-2) were from Biolegend. Antibodies for S1p1 (H-60), CCR8 (c-17), and Ki67 (M-19) were from Santa Cruz. Antibody for IL35/EBI3 was from R&D System.

Single cell suspension was collected from various tissues (hLNs, lungs, and spleens) from sacrificed mice. Surface CD markers were stained at 4°C for 20 min, protected from light. After according treatments, intracellular staining was carried out following surface staining, 4% PFA/PBS fixation and 0.2% TritonX-100 permeabilization. Data were acquired with BD Aria flow cytometer (BD Biosciences, USA) and analyzed by Flowjo (Treestar). Cells were sorted by BD Aria III flow cytometer (BD Biosciences, USA).

Asthma model as previously reported were modified (33). In brief, mice were i.p. sensitized with 100 µg OVA/alum on day −28 and day −14, followed with OVA spray (1 mg/mL in PBS) at days −7, −4, −1 for 30 min/day/mouse. To treat asthma with coimmunization, OVA + pVAX1-OVA (Co-OVA) or Der p 1 + pVAX1-Der-p1 (Co-DERP1) was i.m. immunized at 100 µg + 100 μg dosage (single dose, in right thigh), on day 0 and day 14. To deplete Tregs in vivo, low-dose cyclophosphamide (Cy) was i.v. injected at 20 mg/kg, on day 7 and day 14. To treat asthma with CD25⁻asTregs, Foxp3-eGFP mice (donors) were coimmunized on day −21 and −7. On day 0, GFP⁺ cells were sorted from CD4⁺CD25⁻ population of donor splenocytes, then i.v. transferred into asthmatic mice (recipients) at 3 × 10⁵ asTregs/mouse. On day 14, asthma of recipients was evaluated. Plethysmography was analyzed as reported (33). Briefly, mice were anesthetized and ventilated. Respiratory resistance was stimulated with methylcholine chloride and recorded by AniRes 2005 system (BestLab Technology Co., Beijing, China). Respiratory resistance was analyzed as relative area of peak (R-area) above the baseline before stimulation. To test infiltrating CD4⁺ T cells, lungs were perfused with 10 mL PBS, minced into single-cell suspension and analyzed on flow cytometer. H&E histology was analyzed by counting infiltrated cells in slide image based on nuclei with ImageJ.

All culture experiments were performed in DMEM with 10% FBS, at 37°C 5% CO₂. Sorted asTregs or hLN cells were restimulated with OVA or Der p 1 (10 µg/mL) in the presence of DCs (DC:T = 1:5, 105 T cells/well) for 24 h. Four hours before collection, monensin (BD GolgiStop) and brefeldin A (BD GolgiPlug) were added into the culture to block cytokine secretion. Cells were fixed and permeabilized for intracellular cytokine stain. To evaluate asTreg proliferation, sorted asTregs were labeled with 10 µM eFluor670 (eBioscience) and restimulated with OVA or Der p 1 in the presence of DCs (DC:T = 1:1, 5 × 10⁴ each cells/well) for 3 days, then analyzed as eFluor dilution by flow cytometry. To determine asTregs’ suppression, DO11.10 CD4⁺ T cells were labeled with 10 µM eFluor670, stimulated by OVA (10 µg/mL) with DCs (T:DC = 5:1), then sorted asTregs were added (asTreg:T = 1:5, 5 × 10⁴ asTregs/well). Division was analyzed 3 days later.

To test naïve T cell priming, CD4⁺ T cells from naive DO11.10 mice were purified and i.v. transferred into asTreg recipient mice on day 7 at 5 × 10⁶/mouse. On day 10, hLNs were collected and KJ1-26 cells were stained for maturation and polarization. For i.t. transfer of Teffs, CD4⁺ T cell of asthmatic DO11.10 mice was purified with magnetic beads (R&D System). asTreg-transferred asthmatic mice were anesthetized i.p. with 80 mg/kg pentobarbital, and a sterilized cut (~5 mm) was made right above trachea. 10⁷ purified cells were loaded in 50 µl microinjector then intratracheally injected on day 7. After 36 h, hLN cells of recipients were stained and KJ1-26⁺ cells were analyzed.

To overexpress S1p1 (NM_007901.3, 1149bp), HIV-based lentiviral expression system was used (Cell Biolabs). S1p1 gene was synthesized from cDNA and cloned into expression vector pSMPUW-IRES-Bsd. Together with packaging plasmids, pSMPUW-IRES-Bsd-S1p1 was transfected into 293FT cells. Lentiviruses were harvested from the culture supernatant, concentrated and added into asTreg suspensions at virus:cell = 3:1 ratio. After 24 h, S1p1 expression on cells were validated by S1p-dependent chemotaxis. Four hours after transfection with validated batch of virus, S1p1 overexpressed (S1p1^ov) asTregs were i.v. transferred. S1p1 overexpression on Teff followed the same procedure. To knockdown CXCR3 or VLA-4, FIV-based lentivirus coding shRNA was used (System Biosciences). shRNA carried by the pFIV U6/H1 puro vector were cotransfected into 293FT cells with packing plasmids. Lentiviruses were collected and transfected into DO11.10 Teffs at virus:cell = 3:1 ratio. Expression of CXCR3 and VLA-4 was determined by surface stain. IP-10-dependent chemotaxis was checked by transwell migration with CXCR3 shRNA transfected T cells. The shRNA and control sequences were CXCR3 shRNA sense 5′-AAAGTGTGGATGTTGTTCACGCG-3′, antisense 5′-AAAACGCGTGAACAACATCCACA-3′, VLA-4 shRNA sense 5′-AAAGCAATGGATATGTTGATGTA-3′, antisense 5′-AAAATACATCAACATATCCATTG-3′, control shRNA sense 5′-AAAGCTCCGAACGTGTCACGTTT-3′, antisense 5′-AAAAAAACGTGACACGTTCGGAG-3′.

Lymph node null mice were prepared according to a standard protocol (47). Pregnant female BALB/c mice were i.v. injected with 100 µg of purified murine LTβR-Ig on gestation day 12. To confirm the successful LN ablation, offspring mice were sensitized and challenged to induce asthma, and representative mice were studied for the absence of peripheral LN including hLN, popliteal LN, and axillary LN.

To determine asTreg distribution in hLN, hLNs of asTreg-transferred recipient asthmatic mice were collected on day 2 after asTreg transfer and frozen in liquid nitrogen immediately. hLNs were mounted in OCT (Tissue-Tek) and 5 µm cryosection were cut with Cryostat (Leica). Slides were fixed with 4% PFA and stained with anti-B220-PE. Images were acquired using the A1R confocal microscopy and NIS-Element AR software (Nikon). Image analysis was done using ImageJ.

10⁵ asTregs (WT or S1p1^ov) or asthmatic DO11.10 Teff (WT, CXCR3 shRNA or S1p1ov) were added in upper chamber of Transwell 96-well plate (Corning). Medium or 20 nM chemokine S1p (Cayman Chemical) or IP-10 (R&D systems) were added into lower chamber, respectively. Cells were allowed to migrate for 4 h at 37°C. The number of migrated cells was assessed by FACS with 10⁵ calibration beads (Sepherotech) as internal control. Results are expressed as Migration Index = fold increase of specific migration over unspecific migration without chemoattractant.

Data were showed as mean ± SD. Data were analyzed using one-tailed Mann–Whitney U test between two groups (Figures 1E,G,H, 2B–E, 3C,D, 4D,F, 5B–F, 6B–E, and 7B,D; Figures S1C,F and S3C in Supplementary Material) and two-way ANOVA for multiple dimensional comparison (Figures 1B,C, 2H,I, 3B, and 4C,G,H). Differences were considered to be statistically significant with *p < 0.05 and **p < 0.01.